tools/msgfmt.py

   1 #! /usr/bin/env python
   2 # -*- coding: iso-8859-1 -*-
   3 # Written by Martin v. Löwis <loewis@informatik.hu-berlin.de>
   4
   5 """Generate binary message catalog from textual translation description.
   6
   7 This program converts a textual Uniforum-style message catalog (.po file) into
   8 a binary GNU catalog (.mo file).  This is essentially the same function as the
   9 GNU msgfmt program, however, it is a simpler implementation.
  10
  11 Usage: msgfmt.py [OPTIONS] filename.po
  12
  13 Options:
  14     -o file
  15     --output-file=file
  16         Specify the output file to write to.  If omitted, output will go to a
  17         file named filename.mo (based off the input file name).
  18
  19     -h
  20     --help
  21         Print this message and exit.
  22
  23     -V
  24     --version
  25         Display version information and exit.
  26 """
  27
  28 import sys
  29 import os
  30 import getopt
  31 import struct
  32 import array
  33
  34 __version__ = "1.1"
  35
  36 def usage(code, msg=''):
  37     print >> sys.stderr, __doc__
  38     if msg:
  39         print >> sys.stderr, msg
  40     sys.exit(code)
  41
  42 def add(id, str, fuzzy):
  43     "Add a non-fuzzy translation to the dictionary."
  44     global MESSAGES
  45     if not fuzzy and str:
  46         MESSAGES[id] = str
  47
  48 def generate():
  49     "Return the generated output."
  50     global MESSAGES
  51     keys = MESSAGES.keys()
  52     # the keys are sorted in the .mo file
  53     keys.sort()
  54     offsets = []
  55     ids = strs = ''
  56     for id in keys:
  57         # For each string, we need size and file offset.  Each string is NUL
  58         # terminated; the NUL does not count into the size.
  59         offsets.append((len(ids), len(id), len(strs), len(MESSAGES[id])))
  60         ids += id + '\0'
  61         strs += MESSAGES[id] + '\0'
  62     output = ''
  63     # The header is 7 32-bit unsigned integers.  We don't use hash tables, so
  64     # the keys start right after the index tables.
  65     # translated string.
  66     keystart = 7*4+16*len(keys)
  67     # and the values start after the keys
  68     valuestart = keystart + len(ids)
  69     koffsets = []
  70     voffsets = []
  71     # The string table first has the list of keys, then the list of values.
  72     # Each entry has first the size of the string, then the file offset.
  73     for o1, l1, o2, l2 in offsets:
  74         koffsets += [l1, o1+keystart]
  75         voffsets += [l2, o2+valuestart]
  76     offsets = koffsets + voffsets
  77     output = struct.pack("Iiiiiii",
  78                          0x950412deL,       # Magic
  79                          0,                 # Version
  80                          len(keys),         # # of entries
  81                          7*4,               # start of key index
  82                          7*4+len(keys)*8,   # start of value index
  83                          0, 0)              # size and offset of hash table
  84     output += array.array("i", offsets).tostring()
  85     output += ids
  86     output += strs
  87     return output
  88
  89
  90 \f
  91 def make(filename, outfile):
  92     global MESSAGES
  93     MESSAGES = {}
  94     ID = 1
  95     STR = 2
  96
  97     # Compute .mo name from .po name and arguments
  98     if filename.endswith('.po'):
  99         infile = filename
 100     else:
 101         infile = filename + '.po'
 102     if outfile is None:
 103         outfile = os.path.splitext(infile)[0] + '.mo'
 104
 105     try:
 106         lines = open(infile).readlines()
 107     except IOError, msg:
 108         print >> sys.stderr, msg
 109         sys.exit(1)
 110
 111     section = None
 112     fuzzy = 0
 113
 114     # Parse the catalog
 115     lno = 0
 116     for l in lines:
 117         lno += 1
 118         # If we get a comment line after a msgstr, this is a new entry
 119         if l[0] == '#' and section == STR:
 120             add(msgid, msgstr, fuzzy)
 121             section = None
 122             fuzzy = 0
 123         # Record a fuzzy mark
 124         if l[:2] == '#,' and l.count('fuzzy'):
 125             fuzzy = 1
 126         # Skip comments
 127         if l[0] == '#':
 128             continue
 129         # Now we are in a msgid section, output previous section
 130         if l.startswith('msgid'):
 131             if section == STR:
 132                 add(msgid, msgstr, fuzzy)
 133             section = ID
 134             l = l[5:]
 135             msgid = msgstr = ''
 136         # Now we are in a msgstr section
 137         elif l.startswith('msgstr'):
 138             section = STR
 139             l = l[6:]
 140         # Skip empty lines
 141         l = l.strip()
 142         if not l:
 143             continue
 144         # XXX: Does this always follow Python escape semantics?
 145         l = eval(l)
 146         if section == ID:
 147             msgid += l
 148         elif section == STR:
 149             msgstr += l
 150         else:
 151             print >> sys.stderr, 'Syntax error on %s:%d' % (infile, lno), \
 152                   'before:'
 153             print >> sys.stderr, l
 154             sys.exit(1)
 155     # Add last entry
 156     if section == STR:
 157         add(msgid, msgstr, fuzzy)
 158
 159     # Compute output
 160     output = generate()
 161
 162     try:
 163         open(outfile,"wb").write(output)
 164     except IOError,msg:
 165         print >> sys.stderr, msg
 166
 167
 168 \f
 169 def main():
 170     try:
 171         opts, args = getopt.getopt(sys.argv[1:], 'hVo:',
 172                                    ['help', 'version', 'output-file='])
 173     except getopt.error, msg:
 174         usage(1, msg)
 175
 176     outfile = None
 177     # parse options
 178     for opt, arg in opts:
 179         if opt in ('-h', '--help'):
 180             usage(0)
 181         elif opt in ('-V', '--version'):
 182             print >> sys.stderr, "msgfmt.py", __version__
 183             sys.exit(0)
 184         elif opt in ('-o', '--output-file'):
 185             outfile = arg
 186     # do it
 187     if not args:
 188         print >> sys.stderr, 'No input file given'
 189         print >> sys.stderr, "Try `msgfmt --help' for more information."
 190         return
 191
 192     for filename in args:
 193         make(filename, outfile)
 194
 195
 196 if __name__ == '__main__':
 197     main()